I tend to forget how and why we used to do things in the Unix world.

I was busy building a new, old Unix system this weekend and had to partition the main drive. I've gotten used to just using the whole disk on my Linux systems. I mean why not? If I've got a gigabyte or more that I'm pretty sure I'll never fill up, there's no issue.

But in an older, production system, there were no big disk drives. You almost never had enough, much less too much, disk.

So I'm having to remember more about the way it was done in the '80s.

Here are the partitions I usually built:

• /
• /usr
• /tmp
• swap
• /home
• /usr/local
• /usr/spool

I generally didn't build the equivalent of a /stand partition, because there were not a lot of updates to the typical kernel.

Because drives were small, things like the spool or tmp directories would occasionally get large. If they were a part of the root or usr partitions, there was the possibility that the whole system might stop in case of a runaway process generating a lot of data or files. Keeping /tmp on its own partition limited the damage. There were also significant limits on the number of inodes in these old filesystems, so I would always give uucp and netnews their own partitions.

It took thought and experience to get the numbers right. And although it was a chore, it was also satisfying and fun to configure and manage a busy system that ran well over the long term.